Re: [trustees] Future of trustees

[Nima T. <ni...@jc...>]

Hey Andy,

Good to hear from you again, thought you might have been abducted. If I 
can help in any coding/testing let me know, I'd be more than glad to 
help. Doing so will help me write better man pages too, so its all good.

Nima

Andrew Ruder wrote:
> Ok guys,
> 
> Last night I looked at all my local changes that have not been committed 
> to subversion and what really needs to be done to fix Rob and others' 
> existing problems (as well as finish the documentation for Nima).
> 
> Currently my changes include several changes towards making trustees use 
> general 2.6 kernel semantics.  There are several problems with the 
> source code.  For example, there is a single implementation of linked 
> lists in the kernel, but trustees is still using its own.  There are 
> other similar problems where trustees is simply using old-style code 
> instead of 2.6ish code.  This needs to be fixed as it greatly simplifies 
> the code.
> 
> Also, there needs to be a overhaul on how trustees are stored in their 
> data structures.  There needs to be a quick way to find out if the 
> trustees on one directory are the exact same as those on another 
> directory.  Also, there needs to be a -=very=- fast way to determine 
> trustees for a directory.  Try making a 600-deep directory structure 
> someetime and touching some files in there, your system cpu usage will 
> hit the top as trustees tries to process that big of a directory tree. 
> Both of these changes will be necessary for Rob to see what he wants 
> (and to fix several bugs on the SF.net page).
> 
> Also, I am going to do something with the settrustees script.  There may 
> be some sort of settrustees backend which actually sends it, and a perl 
> or python frontend which takes care of 99% of the logic.  I hate writing 
> parsing code in C and I would be able to integrate a lot of Nima's 
> excellent work on his python scripts.
> 
> In any case, I am certain that these changes will require many changes 
> to the code, and changes mean lots of testing.  Since the new release 
> has come out, it took a while until it stabilized; and I still get the 
> occasional oops report.  (Although, it has been running stable as a rock 
> on my machine).
> 
> So what I am considering is just putting out a 3.0 with the one small 
> subversion fix I have and last night I cleaned up the makefiles to work 
> the -=correct=- way.  (Was having some problem installing into the 
> correct location on newer kernels).
> 
> Also with this, I would make the patches to several kernels to allow 
> compiling trustees into the kernel instead of just as a module.
> 
> This would allow a stable release to be out there while these bigger 
> changes are going in.
> 
> What do you think?
> 
> Thanks,
> Andy
> 
> 
> -------------------------------------------------------
> SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
> from IBM. Find simple to follow Roadmaps, straightforward articles,
> informative Webcasts and more! Get everything you need to get up to
> speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
> _______________________________________________
> trustees-general mailing list
> tru...@li...
> https://lists.sourceforge.net/lists/listinfo/trustees-general